Association rules mining (ARM) is one of the most popular tasks of data mining. Although there are many effective algorithms run on binary or discrete-valued data for the problem of ARM, these algorithms cannot run efficiently on data that have numeric-valued attributes. However, in many real-world applications, the data usually consist of numerical values. It is a difficult problem to determine which attributes will be included in the discovered rules; automatically adjust the ranges of the attributes in the most appropriate way; rapidly discover the reduced high-quality rules directly without generating the frequent itemsets ensuring the rules to be comprehensible, surprising, interesting, accurate, and confidential. Furthermore, adjusting all these processes without the need for metrics to be determined a priori for each data set is of great importance in terms of automating this problem. Recently, numerical ARM has been dealt with as a multi-objective problem that best meets different criteria at the same time. In this study, algorithms which consider numerical ARM as a multi-objective optimization problem were examined and the performance analysis of these algorithms was performed for the first time to the best of our knowledge. A comparative analysis of MOPNAR, QAR-CIP-NSGA II, NICGAR, MODENAR, MOEA_Ghosh, and ARMMGA methods in terms of the number of rules, average support, average confidence, average lift, average conviction, average certain factor, average netconf, average yulesQ, and coverage percentage metrics in the real-world data consisting of numerical attributes was performed. The performances these algorithms were tested with single-objective optimization methods for ARM in this study. It is found that MOEA-Ghosh is the most effective multi-objective method in terms of average support and average confidence measures in data sets containing high number records and attributes. The best results in terms of average support value were obtained by MOEA-Ghosh algorithm and the average confidence values were obtained by multi-objective QAR-CIP-NSGAII in data sets containing relatively few records and attributes. Furthermore, it can be concluded that multi-objective algorithms outperformed the single-objective algorithms with respect to average support, lift, certain factor, netconf, and yulesQ metrics.
Numerical association rules mining Multi-objective optimization Data mining
This is a preview of subscription content, log in to check access.
Compliance with ethical standards
Conflicts of interest
The authors declare no conflict of interest.
This article does not contain any studies with human participants performed by any of the authors.
Agrawal R, Imieliński T, Swami A (1993) Mining association rules between sets of items in large databases. ACM Sigmod Rec 22(2):207–216CrossRefGoogle Scholar
Ahn KI, Kim JY (2004) Efficient mining of frequent itemsets and a measure of interest for association rule mining. J Inf Knowl Manag 3(3):245–257CrossRefGoogle Scholar
Fister I Jr, Iglesias A, Galvez A, Del Ser J, Osaba E (2018) Differential evolution for association rule mining using categorical and numerical attributes. In: Yin H, Camacho D, Novais P, Tallón-Ballesteros AJ (eds) International conference on intelligent data engineering and automated learning. Springer, Cham, pp 79–88. https://doi.org/10.1007/978-3-030-03493-1_9CrossRefGoogle Scholar
Fister I, I Fister Jr, Fister D (2019) BatMiner for identifying the characteristics of athletes in training. Computational intelligence in sports. Springer, Cham, pp 201–221CrossRefGoogle Scholar
Mata J, Alvarez JL, Riquelme JC (2001) Mining numeric association rules with genetic algorithms. In: Kůrková V, Neruda R, Kárný M, Steele NC (eds) Artificial neural nets and genetic algorithms. Springer, Vienna, pp 264–267CrossRefGoogle Scholar
Mata J, Alvarez JL, Riquelme JC (2002) Discovering numeric association rules via evolutionary algorithm. Pacific-Asia conference on knowledge discovery and data mining. Springer, Berlin, pp 40–51CrossRefGoogle Scholar
Moslehi P, Bidgoli BM, Nasiri M, Salajegheh A (2011) Multi-objective numeric association rules mining via ant colony optimization for continuous domains without specifying minimum support and minimum confidence. Int J Comput Sci Issues (IJCSI) 8(5):34–41Google Scholar
Piri J, Dey R (2014) Quantitative association rule mining using multi-objective particle swarm optimization. Int J Sci Eng Res 5(10):155–161Google Scholar
Ramaswamy S, Mahajan S, Silberschatz A (1998) On the discovery of interesting patterns in association rules. In: Proceedings of the 24th international conference on very large data bases, California, USA, pp 368–379Google Scholar